Virtual US-RSE Workshop Recap
Published: Jun 3, 2020
Given recent current events around the novel coronavirus outbreak, many individuals and organizations have been forced to adapt. For the US Research Software Engineer Association (US-RSE), this meant the unfortunate postponement of an in-person community building workshop that was to take place April 23-24th, 2020. In order to keep up the community’s strong momentum, we decided to host a virtual workshop that could be attended from the comfort and safety of our own homes. While not aiming to replace the in-person workshop, this virtual workshop format came with some benefits of its own, such as additional participation by those who would not have been able to attend the in-person workshop, reduced environmental impact from not having to travel, and less disruption in the lives of attendees.
The virtual workshop was well attended, with 30-90 participants in each session. The overwhelming majority of participants were affiliated with academic institutions in the United States. To avoid so-called “Zoom fatigue,” the workshop was planned in three sessions of 1.5 hours each, held over two days. The three sessions were called: “The RSE Landscape”, “Technical Talks”, and “Next Steps for the US-RSE Association”, which was a breakout and discussion session. Recordings of session 1 and session 2 are available on Youtube.
Session 1: The RSE Landscape
The first session, chaired by Charles Ferenbaugh of Los Alamos National Laboratory, introduced the audience to the concept of research software engineering. The session started out with a talk by US-RSE Association Chair Ian Cosden (Princeton). He provided a brief history of the formation of the US-RSE Association, including its relationship with similar organizations in other countries, as well as the definition of “Research Software Engineer.” Ian’s talk was followed by talks from Mahmood Shad (Harvard), Christina Maimone (Northwestern), and Nuyun Zhang (Georgia Tech) on the research software engineering groups in their respective institutions. Slides from these talks can be found on the event page, and a recording of session 1 can be found on Youtube.
These talks were followed by a panel discussion on the situation RSEs face in the modern research landscape. Topics discussed included:
- How do you gather metrics to make the case for an RSE group at an institution? Often the most effective metrics are the success stories of faculty members, e.g.: “See this paper? We couldn’t have done it without RSE support.”
- How can we create opportunities for RSEs, who are scattered across different research institutions, to learn together and work together? Some institutions do this by building internal groups for software developers, but this can be hard for decentralized groups. US-RSE wants to provide this on a larger scale, but we’re still figuring out how. Technical talks, Slack channels, and local/regional meetups could be part of this.
- How are institutional RSE groups funded? One current model is a centrally-funded group of RSEs in the IT department, with projects that hire RSEs paying back into the IT budget. Others live under CS departments or in high performance computing (HPC) groups. We as a community need to gather an answer for this, and help provide support to RSEs who are funded through “soft money” and wonder what will happen at the end of each project.
- Can we build a library of RSE use cases and success stories? We should definitely try to do this. If people think this would be useful, or if they have stories to contribute, let us know!
- What are successful recruitment and retention strategies for RSEs? This isn’t easy in academia, given the competition from industry. Working with HR can close part, but not all, of the salary gap compared to industry. We can stress job stability, work/life balance, opportunities for learning, and contribution to society as intangible benefits.
- How do you identify the best candidates for RSE positions? Strategies can include discussing past projects, code samples, take-home code challenges, whiteboard sessions, technical questions, and behavioral questions.
- Can RSEs operate as co-researchers on projects? Panelists gave several examples of projects where RSEs were embedded long-term, acquiring some domain knowledge, contributing to research, and contributing as co-PIs.
Session 2: Technical Talks
The second session was chaired by Chris Hill of MIT. It featured seven short technical talks by speakers from the US-RSE community, giving a sampling of the wide range of work done by RSEs.
- Data Analytics and Provenance Tools Development at GT PACE - Fang (Cherry) Liu, Georgia Tech
- Development of ASPIRE Python Package: Reconstructing 3D Density Map of Biomolecule From 2D Cryo-EM - Junchao Xia, Princeton
- Research Data Management for Medical Data with Pyradigm - Pradeep Raamana, Baycrest Health Sciences
- Development of an Automated High-Throughput Animal Training Platform - Sarah Leinicke, Harvard
- Software for Renewable Energy Integration - Kaspar Mueller, Intellectual Ventures
- Spack: What Is It, What’s New, and What’s on the Roadmap? - Todd Gamblin, LLNL
- Scientific Software Management at PACE - Kevin Manalo and Chris Stone, Georgia Tech
Slides from these talks can be found on the event page, and a recording of session 2 can be found on Youtube.
Session 3: Next Steps for the US-RSE Association
The final session, “Next Steps for the US-RSE Association,” was chaired by Jordan Perr-Sauer of the National Renewable Energy Laboratory and facilitated with breakout rooms by Sandra Gesing of the University of Notre Dame. This session utilized breakout rooms in Zoom to provide a spontaneous, discussion-based environment. The main goal for this session was for participants to “get to know each other.” Our format called for three breakout sessions, structured such that each participant was randomly assigned to a breakout room with 4-6 other participants. In this way, each participant would introduce themselves and hear introductions from 12-18 other participants over the course of the session. Each breakout session lasted 15 minutes and was centered around a discussion topic. A suggested agenda was provided to the participants using Google Drive. After each breakout session, participants were sent back to the main room and encouraged to record their thoughts from the discussion in the Zoom chat.
This format was new for the organizers, and certainly experimental, but it turned out to be successful and was well received by participants who enjoyed the highly interactive session. Reshuffling participants into new breakout rooms allowed them to meet many new people, but because the introductions were repeated each session, the time for focused discussion topics ran out sooner than anyone would have liked. The discussion in the breakout rooms were so engaging that participants universally wished there was more time. The goal of meeting new people within the community and getting to know each other was a clear success. However, to allow more time for discussion possible modifications for future workshops include: longer breakouts, smaller breakout rooms, and/or consistent breakout rooms for sessions.
The breakout sessions focused on the following predetermined discussion themes:
- What are some problems that US-RSE can help solve?
- What activities should US-RSE encourage or participate in?
- How to establish an RSE group at your own institution.
While it’s impossible to capture all the discussions, a few themes emerged that we summarize here. In both of the first two sessions, multiple breakout groups touched on various aspects of RSE career path, professional development, hosting events aimed at RSEs, and developing resources for organizing groups of RSEs.
Participants wanted to see US-RSE help establish and legitimize a career path for RSEs. The role of RSEs within an organization may be poorly understood when compared to more traditional research roles. It would be helpful to have a national organization like US-RSE, rather than local institutions, provide formal definitions of various RSE positions. Providing materials documenting the value of RSEs within a research institution could help RSEs better make their case to management. One specific suggestion that came up was to create a catalog of the different business models in use today by RSE groups.
Additionally, participants called for community building and knowledge sharing activities such as workshops, webinars, training courses, and working groups. The creation of new conferences (such as this virtual workshop), networking events, and user groups was suggested to help build the RSE community. Such groups might operate on a local level, connecting RSEs between institutions in the same area.
For the third session, the organizers decided to have one large session rather than using breakout rooms. This session utilized a Q&A format, where participants asked questions through the Zoom chat. Questions centered around the funding and business models in use by RSE groups, as well as how to raise interest for informal groups of RSEs at an organization.
Although this virtual workshop was organized quickly in the midst of a global pandemic, we view it as a resounding success. The various session formats allowed for different kinds of participation, whether the attendee wanted to just be a fly on the wall and watch talks, or be an active participant by giving a talk or joining a breakout session. This format proved to work remarkably well, and we intend to organize more virtual workshops and other online events in the future. If you missed this virtual workshop, you can find slides and recordings on the event page.
Check out upcoming events on the US-RSE calendar, or join the US-RSE Slack to join the conversation and share your ideas. If you’d like to help organize an upcoming virtual workshop or event, reach out to @sc on slack. Please join us at PEARC20, SC20, and the in-person community building workshop (whenever it is able to be rescheduled). More details on these workshops, including announcements of opportunities to actively contribute to the programs, will be announced soon.